Font Size

100% 150% 200%

Background Colour

Default Contrast
Close Reset

Roxburgh Castle

Castle (Medieval)

Site Name Roxburgh Castle

Classification Castle (Medieval)

Alternative Name(s) Old Roxburgh Castle; Protector Somerset's Camp; Marchmount

Canmore ID 58412

Site Number NT73SW 12

NGR NT 71307 33729

NGR Description Centred NT 71307 37729

Datum OSGB36 - NGR


Ordnance Survey licence number 100057073. All rights reserved.
Canmore Disclaimer. © Copyright and database right 2020.

Toggle Aerial | View on large map

Digital Images

First 100 images shown. See the Collections panel (below) for a link to all digital images.

Administrative Areas

  • Council Scottish Borders, The
  • Parish Roxburgh
  • Former Region Borders
  • Former District Roxburgh
  • Former County Roxburghshire

Accessing Scotland's Past Project

Roxburgh Castle, also later known as Marchmont, occupied a triangular hill flanked to the north and south by the Rivers Tweed and Teviot. Throughout its occupation the castle changed hands many times, reflecting the troubled history of the Borders as a whole. In the sixteenth century the Scots utterly destroyed Roxburgh Castle to prevent it being used again by the English.

The castle first appears on record in 1125, and from its earliest days it was an important centre of power at regional and national level and was often involved in key events in Scotland's history. In the Wars of Independence, the castle changed hands more than once, and in 1313 was famously taken by Sir James Douglas for the Scots who, with only 60 men, scaled the castle mount and put many of the unsuspecting garrison to the sword. By 1356, however, the castle had once more fallen to the English, who held both castle and town until 1460 when James II laid siege to the fortress with an impressive army drawn from across Scotland. The Scots, fearful that English forces might occupy it, finally destroyed the fortress in the sixteenth century. In the 1550s, English forces briefly re-occupied the site with the intention of rebuilding a fortress there.

Aerial photographs show that two ditches defended the approach to the castle between the Rivers Teviot and Tweed. Some fragmentary sections of curtain walling can be seen along the southern side of the promontory, facing the Teviot. These would have enclosed the whole area. Throughout its long history, the castle was increasingly fortified, especially during the century of English occupation. Two charters survive, describing in great detail the work that was done to the castle, giving a good account of how it looked by the fifteenth century when James II besieged it.

Text prepared by RCAHMS as part of the Accessing Scotland's Past project

Archaeology Notes

NT73SW 12 71307 33729.

(Centerd NT 7131 3373) Roxburgh Castle (NR) (Ruins). Protector Somerset's Camp (NR).

(Centred NT 7135 3353) Cauld (NR)

OS 6" map, Roxburghshire, (1918-38).

Roxburgh Castle. Marchmount, the castle of Roxburgh, once the strongest fortress on the whole Border but now almost wholly demolished, lies about three-quarters of a mile W of Kelso immediately above the junction of Teviot and Tweed, these streams flanking the site on two sides. The mount itself is a kaim, 70ft to 80ft high, with a summit roughly triangular in shape, pointing E (Calendar of Documents, J Bain ed. 1881), and measuring 800ft in length by 350ft across the base. The Teviot, which washes the foot of the S escarpment, renders other defences unnecessary on that side ; but along the N side a great ditch, still 12ft in average depth, shuts off the haughland lying between the mount and the Tweed. This ditch stops at the apex of the site where there seems to have been a foretower with another ditch in front of it to bar approach from the E. At the W end of the position a gully, which separated the mount from the ridge ending in the Gallows Knowe, has been extended, deepened and provided with a rampart on its counterscarp. Where this re-entrant dies out on the bank of Teviot the remains of an old dam, or "cauld", span the river. There is a local tradition that the purpose of the dam was to provide water for flooding the hollow; the levels now seems to be unsuitable, but none the less in 1545 Bartholomew Butler, who when York Herald accompanied the Earl of Hertford on his invasion of Scotland in that year, thought it a practicable proposal. (D Laing 1855). The "fosse" of the castle, by which is meant the great ditch, was excavated in 1400, (Calendar of Documents, J Bain ed. 1888), the year in which the re-entrant was extended. (Pell Records). A general view of the site is given in the air photograph reproduced as RCAHMS 1956, fig. 519.

The early castle was an enclosure containing, among other buildings, the church of St John. Both castle and church come on record in a charter granted by David I about 1128 (A C Lawrie 1905). The only other building of importance existing at this time was the tower, or donjon. In this tower David I imprisoned Malcolm, brother of the Mormaer of Moray, in 1134, (A O Anderson and M O Anderson 1936). Either in the same tower, or more probably in its successor, the Gascon governor, in February 1313-14, held out for a day after the rest of the castle had fallen by escalade to Sir James Douglas. A vivid picture of this exploit is given by Barbour, who describes the building as "the gret toure", (J Barbour 1894) a description confirmed by "la graunt tour" of Sir Thomas Gray (Scalacronica 1836). The tower in question is possibly to be identified with "le Doungyon vocatum Douglase Toure" mentioned below.

There is nothing now to be see of tower or church. As the castle was continuously occupied during the four and a half centuries previous to its final destruction in 1550, and was alternately held and besieged by the forces of the two kingdoms, there must necessarily have been much repair and alteration, only a few particulars of which have been recorded (Calendar of Documents, J Bain ed. 1888; Rotuli Scotiae 1819;

L F Salzman 1952). But an account of the castle as it stood in 1416, when in the hands of Henry V, is preserved in the Public Record Office, London. (W R Cunningham 1923). . . . .

The place was besieged by the Scots in 1417 before major repairs could be effected, and men and munitions had to be hastily brought in. Artillery came from London- not the first time that the new arm had been seen here, as between 1382 and 1384 "gunpoudre" and "artellar" had been supplied to the castle (Calendar of Documents, J Bain 1888). In 1419, however, Henry V ordered Robert Fekenham, his master mason, to engage masons and carry out repairs forthwith; (Rotuli Scotiae 1819), and no doubt some of the defective timberwork mentioned in the survey had already been temporarily repaired. The accounts for repairing the walls were settled in 1419, and in the following year 500 stones of iron were purchased to make gates for the castle(Calendar of Documents, J Bain ed. 1888). In 1460, having been continuously in English hands for over a century, the castle sustained its most memorable assault. On Sunday, 3 August, "King James II, with ane gret oist, was at the sege of Roxburgh" (Auchinleck Chronicle nd). It was at this siege that King James II was killed. The Scots carried the castle by storm and it was "doung to the ground". The condition of the place thereafter may be inferred from a charter of 1488 in which James IV grants to Walter Ker of Cessford "castrum et locum castri vocatum le Castelsted" (Reg Magni Sig Reg Scot 1424-1513). This castlestead or site recurs in charters of 1500 and 1542. In 1545 the Earl of Hertford examined "the castell of old Rockesburghe, being within a quarter of a myle of Kelso" on Henry VIII's instructions, and reported that it was "altogither ruyned and fallen downe...(State Papers, Henry VIII 1830-52). "The building of the fortalice of Roxburghe" is mentioned in the arraignment of a Scottish traitor in October 1547 (R Pitcairn 1833). On the 15th of the same month Lord Grey of Wilton reports to Somerset that the work is well forward and sends him a "platt" or plan prepared by the surveyor, (Scottish Papers, J Bain ed. 1881), whose name was Rydgewey (W Douglas 1921). By rare good fortune this "platt" has been preserved in the library of Belvoir Castle, and it is reproduced here (RCAHMS 1956, fig.517). . . .

It is clear that this new fort covered only the western and higher part of the old site, about a third of which remained unoccupied. The plan is admirably drawn to scale, and a legend in the middle of the drawing states that "This Platte Contaynyth xl fotte to the inch". At top and bottom are shown the two rivers, the upper one labelled "The Watter of Tweed", the lower one, labelled "watter" at each end, is identified below as "The Watter off Tyvytt". The fort depicted is an oblong enclosure with the major axis running E and W, the SW corner containing the entrance. Towards the lower left-hand corner of the plan may be seen the beginning of the re-entrant referred to above. On the scarp of this re-entrant there is an external enclosure, evidently part of the older castle and not of the fort, containing "The Welle" at its SE corner (supra). The E side of the enclosure is bounded by "A Dyche", the outer work of the fort on this side. This ditch runs from the N edge of the mount to a bastion protecting the entrance of the fort, which, as stated above, lies at the SW corner of the enclosure and is here called "Ye gaatte". In the angle formed by the scarp of the ditch and the N side of the enclosure is a second bastion labelled "The Bell Mowntte"; as early as 1411 the castle contained "a great bell called 'Watchbelle' value ?40" stg., (Calendar of Documents, J Bain ed. 1881), from which

no doubt this mount took its name. The two bastions are connected by a curtain running parallel to the ditch and provided with a wall-walk on its inner side. In the base of the one beside the gate an angled passage leads to a little square chamber containing a gun-port which commands the entire length of the ditch. The Bell Mount is reached by flights of steps on S and E.

The gate of the fort opens into "The Base Cowrtte". This court widens towards its inner end where a doorway admits to the main enclosure. On the E it abuts on "A Mount". This mount is reached from the enclosure by a forestair, and in its base there is an embrasure, containing a gun-port from which the base-court can be raked with fire. On the N side of the main enclosure an angled curtain, interrupted midway in its length by a projecting tower which contains a gun-port facing W, runs from the Bell Mount to "A Mowntt" at the NE corner. This latter mount, reached from the S by a flight of steps, communicates with a little tower projecting N, in which there is a gun-port facing E. The E side of the enclosure is also protected by "A Deiche" and a curtain on the scarp of this ditch extends from the mount last mentioned to the SE corner, where it juts out to the E before returning along the S of the fort. In this projection there are two gun-ports facing along the ditch. Behind this E curtain and also behind the W part of the N curtain there are platforms for guns. Behind the remainder of the curtains there are wall-walks. The curtain on the S is interrupted midway by "Ye Captyns Logynge". This building, which is entered from the main enclosure, abuts on the mount beside the base-court. It is entered through the well of a projecting stair-tower and contains two rooms on the ground floor. The one first entered, the bigger of the two, has a large kitchen-fireplace on the W and its E end is partitioned off. A doorway at the NW corner admits to the inner chamber, which has a fireplace in the W gable. On the E of this "lodging" and abutting on the S curtain is a second building, "The Brewe howse (Calendar of Documents, J Bain 1888). Evidently a wooden building, for the "frame" was brought from Wark. (Calendar of Documents, J Bain ed. 1888), and Backe howse". This has two doorways in the side next the enclosure and an oven in the E gable. Against the gun-platform on the E stands "The Store hows", a long range divided up by partitions, lit and entered from the W. and having a large fireplace in the N gable. Against the gun-platform and wall-walk on the N there are two sets of "Logyngis". A third building, which surrounds a little triangular court, is not given a name.

As a military work this new fortalice had but a short life. In terms of the peace treaty concluded between France and England at Boulogne in 1550, Edward VI evacuated Scotland and demolished the forts that he had erected at Dunglass, Eyemouth, and Roxburgh. And in the sequel, Mary, Queen of Scots, in 1533 granted "terras et baroniam de Auld Roxburgh cum castri loco", etc. to Andrew Ker (Reg Magni Sig Reg Scot 1546-1580).

A plan of the castlestead as it now appears is given in RCAHMS 1956, fig.514. The E and W gates of 1416 (supra) can readily be located at either end of the site. At the former may be seen foundations of a foretower and at the latter there are fragments of a gate-house. There are considerable remains of the curtain that ran between the two along Teviotside, and in some places it still stands to a maximum height of 13ft- this is all that remains of the wall that was rebuilt to a height of 30ft in 1378. The two posterns of 1416 can still be identified in the curtain and, farther E, the "round tower towards Teviotside". This last was probably D-shaped and had a vaulted basement. Its masonry is regularly coursed and squared and is built with pinnings, contrasting with the rubble masonry of the curtain. On the outer side of the curtain provision for timber bratticing can still be seen. Of the N curtain and its central tower only two inconsiderable fragments are left. On the W side of the one to the W. there is a mound standing about 12ft higher than the reminder of the site ; this is obviously the "Bell Mount" of 1547, and its position suggests that it is the site of the donjon called the Douglas Tower. If this is the case it may also be the site of the "great tower" of 1313-14, as well as of the tower of 1134 (supra). On the W of the mound runs the ditch of 1547, and in front of that lies the enclosure that contained the well. The two fragments of masonry at the S end of the ditch are probably remains of the mount beside the gate. The E ditch of 1547 has disappeared ; there is some trace however, of the curtain in front of which it ran. Illustrations of some of the foregoing features are given in RCAHMS 1956, figs.520, 521, 522, 523, 524, 528 and 529.

There is no visible evidence for the tradition of 1649 (W Macfarlane 1906-8), that there were "vaults underground that went to both... Rivers".

RCAHMS 1956, visited 14 June 1932.

Roxburgh Castle is as described above.

Revised at 1/2500.

Visited by OS (RDL) 4 December 1963.

This site was visited in the course of fieldwork by Dr. T.C. Welsh in 2000

for further details see MS/767/6

Architecture Notes



RHP 6075 Mid-16th Century, Belvoir MS - photostat copies of plans

Non-Guardianship Sites Plan Collection, DC28502-DC28504, 1964, 1979 & 1981.


Online Gallery (1306 - 1329)

The year 2014 sees the 700th anniversary of the Battle of Bannockburn, in which the army of Robert I of Scotland defeated that of Edward II of England. The battle marked a major turning point in the long, drawn-out struggle of the Wars of Independence.

The Wars have had a lasting influence upon all the nations of the United Kingdom and upon the national story. Each age has seen fit to commemorate the events in its own way: through the perpetuation of the genuine historical associations of buildings and places and also through the endowment of others with improbable or fanciful traditions. Where past generations allowed its historic buildings to decay and disappear, later generations began to value and actively preserve these for their associations. Where an event lacked a tangible reminder, as at Kinghorn where Alexander III was killed in a riding accident, a commemorative monument would be erected to act as a focus. The Wars of Independence predate the fashion for accurate portraiture: the weathered, generic military effigy of Sir James Douglas is one of the few to survive in Scotland. Later centuries saw a need and supplied it by a crowd of images of its historic heroes, William Wallace and Robert the Bruce, each depicted according to contemporary taste and imagination. The opening of the new heritage centre at Bannockburn takes this into a new dimension, through the use of three-dimensional, digital technology.

RCAHMS Collections hold many images of these buildings and locations from battlefields, castles and churches, to the many commemorative monuments erected in later years. This gallery highlights a selection of these, including antiquarian sketches, photographic and drawn surveys, and architectural designs.

Aerial Photography (3 July 2003)

Project (2006 - 13 March 2009)

This project was initiated after the Time Team excavations of 2003 and the formation of a small research group to stimulate research on Roxburgh. A conference was held at Stirling University in 2005 to review what was known about the site. In the light of this RCAHMS decided to carry out a new topographical survey of the site, providing a 3D terrain model as the base for a GIS combining all the known data from excavation and survey. This is intended to provide a springboard for future research. Four new datasets were created in the course of the RCAHMS survey 2006-2009; aerial transcription data from aerial photographs taken in 2006 and 1984, a Lidar dataset of the site and surrounding area, a terrain model based on a differential GPS survey and an earthwork survey. The terrain model and the earthwork survey cover the castle and burgh, but cropmarks have only been recorded on the photographs of the E part of the burgh. The Lidar data covers an area of about 4.5km from E to W by 2.5km transversely around and including the site itself. This data has been combined with information from estate plans and documentary sources to provide an interpretation of the layout of the street pattern of Roxburgh and the development of the castle.

Field Visit (2006 - 13 March 2009)

This topographic survey of the castle has revealed more of the later artillery fort and earlier castle than was known hitherto; it was aided by a three dimensional model based on ground survey with differential GPS and observations made during a detailed hachure survey. The plan of the artillery fort of AD 1547 published in the Inventory (RCAHMS 1957, Fig. 517) was overlaid in GIS to help disentangle its features from the earlier ones, and it was also possible to use the 1416 survey of the state of the castle, quoted in the Inventory (RCAHMS 1957, 408), to aid interpretation of some of the earlier features.


The castle occupies the summit of a glacial ridge that rises to c. 30m in height and extends over 320m from NNE to SSW by 140m transversely. The ridge is broken by natural terraces and gullies and the main court of the castle occupies the highest part which is bounded by steep slopes to the Teviot on the S and to the haugh beside the Tweed on the N, dominating both and providing a strong natural defence of the narrow neck of land between the two rivers. This high mound was improved as a defence by the construction of a broad ditch that runs along its base on the NW and SSW arcs, but is absent on the SE where the Teviot washes the foot of the mound too closely for a ditch to be needed. An outer ditch was excavated to the SSW of a detached part of the ridge to the S that runs from WNW to ESE, the prominent WNW end of which is known as Gallows Knowe (a trench dug into the summit was used as a Home Guard defence point), but there is no clear trace of defensive or structural features that would suggest this was an outwork of the castle. Access to the castle mound is restricted to the NE tip and SE corner of the ridge since elsewhere the slopes are too steep to climb easily, and it is here that the entrances are to be found. The earliest structure occupies the summit of the mound, enclosing a triangular area extending some 280m from the SW corner to the entrance at the NE tip by 60m transversely, an area of c. 1.4ha. Apart from a lower terrace at the SW end, up to 40m in breadth, the summit is more or less level, but is divided into three parts by the artillery fort that occupies the central part of the castle mound. The secondary features relating to the fort and its internal structures will be described below.

Despite the disturbance of the castle by a later artillery fort (below), it is possible to see sub-divisions into wards or courts for defensive reasons at both ends of the castle. At the SSW there is an outer curtain on the lower terrace that runs up to the SW gate. This may form an outer court on the terrace if the robbed wall on the higher lip of the mound at this end forms a complete inner curtain. Indeed, the footing of a wall that runs from the passage of the S entrance (NT 71322 33670) upslope NW towards the return of this inner curtain may serve to complete the circuit, but is unattached at the entrance passage, unless traces of an inner lining of the passage can be found. At the NE there is an outer court where the curtain on the NW extends beyond the large D-shaped tower and building that runs across the full breadth of the mound to form an inner defence line, creating a small triangular outer court at that end between it and the outer gate and barbican. The outer curtain on the SE of this outer court is only indicated by a raggle of a return wall on the SE of the SW end of the surviving side of the barbican. It is presumed that this wall duly turned SW at the lip of the mound to run down to the D-shaped tower. The NW wall of the barbican has been completely robbed leaving a platform that expands at the SW to form what may be a base for a turret.

The curtain walls

The primary walls of the castle are best-preserved on the SE where a curtain wall of coursed, lime-mortared, roughly-dressed masonry with pinnings, 1.8m in thickness, still stands to as much as 5m in height in places. There are at least two postern gates in the surviving masonry along this wall, one near the inner entrance on the S (NT 71336 33695), and the other just S of a slight turn in the curtain wall (NT 7136 3375). Both of these entrances are described as requiring iron gates in the 1416 survey (RCAHMS 1957, 408). Along the NW arc immediately S of the NE entrance the curtain has been reduced to a robber-trench visible as a terrace about 2m wide, or has been subsumed in the later fort rampart further S, although a break of slope at the foot of the rampart appears to mark its line. One fragment of original curtain with the same character as the SE curtain still survives on the line of the robber trench and in front of the artillery fort rampart (NT 71282 33729). It retains traces of openings that suggest it was once part of a building, possibly part of the Bell Mount of the 1547 plan, or the Douglas Tower of 1416. The curtain S of the NE entrance on the SE has also been robbed as far as the artillery fort, leaving no trace except for a break of slope. On the SSW where it runs along the edge of the lower terrace it is visible as a grass-covered footing, about 1.8m in thickness and up to 0.5m in height. This part of the curtain runs up to the W of the S gate and is overlain by a second line of curtain wall, also visible as a grass-covered bank, that runs along the back of the same terrace. The primary curtain on the NW of this lower terrace continues as a footing until the junction with the second phase of curtain at the back of the terrace from whence it is reduced to a robber trench as it extends NE until it too is overlain by the artillery fort. A large round platform some 11m in diameter is situated just outside the curtain at this junction with the second phase of curtain which the robber trench appears to cut across. This is one of five round or semi-circular platforms along the line of the NW curtain wall that may mark the location of towers. The best defined, also measuring c.11m in diameter, is situated 50m to the NE (NT 71267 33707) and cuts into the robber trench of the curtain suggesting that the tower has either been built onto an existing wall or is coeval with it. A third is situated at a turn in the NW curtain to the NE (NT 71324 33810) and there are two other platforms equally spaced along this northerly sector of the wall. In front of the SSW outer curtain, there are two interdigitated groups of three terraces, the one subrectangular immediately adjacent to the wall and the outer line, circular on plan (c.7m in diameter) and set about 1m in front of the curtain. These may indicate the location of turrets at different stages in the history of the castle. At the edge of the summit of the castle mound a terrace up to 4m in breadth marks the line of the inner curtain wall. This has been completely robbed except for the stub of a narrower wall on the SW end of the blocking wall on the SE end of artillery fort ditch (see below). This wall is 2m thick and formerly turned a sharp corner to merge with the NW curtain on the W, a junction that is no longer clear due to robbing. The site of a well that is depicted on the 1547 plan of the fort is marked by a semi-circular scar in the outer edge of the artillery fort ditch at NT 7128 3368 (see below). This is presumably the castle well, since no other is evident.

The D-shaped tower and NE entrance

At the NE end of the castle court there is a very substantial circular tower, c.14m in overall diameter, with squared ashlar walls 4m in thickness, still standing some 3.5m in height. The raggle of a wall shows that it was squared at the back and another at right angles indicates it extended to another compartment behind. The projecting D-shaped tower had a vaulted basement and must have stood several storeys in height. It is bonded to a length of curtain, 3m in thickness and 8m in length that runs from NW to SE between the tower and the edge of the castle mound where a return with the curtain may be inferred, but has been robbed. Between the tower and the NW curtain there was a space of about 5m, wide enough for a gate of which there is now no trace. The terrace at the rear of the surviving structure of the tower and curtain is 6.5m in breadth providing space for a building along the back. It is possible this is the ‘round tower towards Teviotside’ described in the 1416 survey quoted in the Inventory (RCAHMS 1957, 408). In front of this structure lies the NE outer gate in the form of a barbican, of which only the SE side remains, extending to 8m in length and standing to about 3m in height with walls 2.5m in thickness. The stop for a door is still visible at the foot of the wall on the NW side. Robber scars in front suggest that the barbican extended another 10m in front of this surviving fragment. Such is the steepness of the slope in front of the gate, it would seem that steps down would be the only practical method accessing the entrance at this end, similar to that on the motte at Sandal Castle (Butler 1991).

The S entrance

At the S corner of the castle mound there are fragments of the second entrance which comprises two thick lumps of masonry on either side of a passage. Some squared ashlar masonry survives on the NW preserving the grooved, rotational wear scar of an iron yett. The footings of the walls exceed 3.5m in thickness with a central passage 3.35m across. In effect this was a barbican enclosing a passage that led to an inner gate 33m to the NE, protecting access to the court. This latter gate survives as a return to the NW on the SW end of the SE curtain, but the wall thickness is just the same as that of the curtain, which does not indicate a heavily defended entrance at this point. The springing for the start of a barrel vault is visible here, suggesting a covered passage, but this might be from a reworking of the gate in the 16th century (see below). A break of slope that runs down towards the ditch in front the gate suggests a ramp had been constructed up to the entrance. As the ditch comes to a butt end here, no bridge would seem to have been required.

Other buildings

Only one possible building was observed within the court of the castle, lying on the inside of the NW curtain a short distance W from the round tower at the NE end. This building is visible as grass-covered, stone footings , measuring internally about 9.5m in length by 4m in breadth (NT 71368 33833). A 15m wide terrace along the inside of the SE curtain S of the D-shaped tower may be the location of buildings of which no other structural evidence remains. All other buildings are described in the context of the artillery fort, including the Captain’s Lodging of 1547 which may be a reused castle building (see below).

The great ditch

A ditch, which varies in breadth from 10m to 20m, has been cut into the base of the mound on all sides, except the SE where the steep slopes drop directly down to the bank of the River Teviot. On the NW it has been reduced by the building of a plantation dyke in the post-medieval period and further filled in to construct the public road in the late18th or early 19th centuries The up-cast from the ditch is visible as mounds at the NE end and at the NW and SE ends of the SSW arm of the ditch. At the NE end a causeway has been built across the ditch to make access easier, probably for robbing stone from the castle.

The 16th century fort

The fort comprises an enclosure extending c. 110m from NE to SW by 65m transversely, built across the middle section of the castle. The interpretation of this fort is aided by a plan made in 1547, preserved in the papers of the Earl of Rutland at Belvoir Castle and published in the Inventory (RCAHMS 1957, Fig 517).

The walls consist mainly of earthen ramparts retained by a stone wall, of which fragments survive on the NW. In places the stonework of the castle was reused, with a surviving fragment towards the SW where the Bell Mount was situated (described above), but on the SE the fort reused the castle’s curtain wall without any rampart at all, except for a mound referred to as ‘A Mownt’ on the plan near the S entrance. The earthen ramparts are up to 20m in thickness on the NE and SW, but only 13.5m on the NW, suggesting variations in height. A ditch 20m in breadth is visible externally on the SW, and this has a stone wall 3.2m in thickness and up to 3m in height across its SE end. The collapse of the rampart following the robbing of the casement walls on the NE probably accounts for the lack of any visible sign of ditch on this side. A bastion shown on the 16th century plan at the NE corner may be indicated in a change of stonework of the curtain from the earlier coursed masonry of the castle to thinner rough coursed work. The gap at this corner in the rampart may be due to the need for access to the central court of the fort in robbing the site after its abandonment. A passage through the middle of the rampart on the NW may be traced on the ground as a hollow that provides access to a turret and is indicated on the plan. Fragments of mortared masonry are evident on both sides of the passage here. The Bell Mount, shown on the plan with steps up from the SW rampart, is also the location of the highest earthwork mound on the site. A less prominent mound on the NW end of the NE rampart is also indicated on plan as the location of a turret with stairs up to it from the rampart. The only entrance is from the S, reusing the old inner gateway of the castle at that end. It is suggested that the springing of a barrel vault visible in the stonework of the surviving pillar of the entrance belongs to the fort.

Fort buildings

The various platforms aligned on the same axes as the fort walls inside the enclosure suggest that they are coeval with the fort. Indeed, the site of a long range marked on the fort plan as ‘The Store House’ immediately inside the NE rampart is indicated on the ground by a rectangular platform set into the slope on the NW, measuring 30m from NW to SE by 6m transversely. Of two platforms for ‘lodgings’ shown on the plan along the back of the NW rampart, that to the NE is indicated by a terrace. Others, including ‘the Brew House and Back House’ (ie. bake house) immediately NE of ‘the Captain’s Lodging’, are visible as earthworks on the SE. The Captains’s Lodging itself is marked by a large oblong hollow, about 24m by 12m in extent, and a gap in the curtain wall of the castle, there being no fort rampart on this side. Some stonework of the NE wall may still be traced but otherwise the robbing has been comprehensive. At the NE end of the hollow, there is a raised terrace about 5m across that marks a subdivision of the building into two unequal compartments. This does not match the plan, which shows a narrow room at the NE, a large room in the middle with a large fireplace and a smaller room with a fireplace at the SW end. The hollow suggests a basement for what was probably the most impressive of the buildings, but the features on the map are likely to be the upper floor of the building, which may well have had a basement and at least one upper floor. An oblong mound along the NW edge of the hollow is marked on the plan as a stair. The stonework of the curtain beside the building is now largely robbed but a clear butt joint is visible, indicating that the Captain's Lodging post-dates the curtain. It is possible that this is a reworked building from the castle.

Note (3 November 2015)

This monument was delisted and was removed from the list of buildings of special architectural or historic interest.

The structure is still designated as a scheduled monument under the Ancient Monuments and Archaeological Areas Act 1979.

Information from Historic Environment Scotland, 3rd November 2015

Note (July 2017)

The burgh

Across the river Tweed from Kelso, there is a grassy field used for point-to-point racing. At its west end where the rivers Tweed and Teviot narrow to an isthmus 150m wide, there is a 100 foot high mound grown with oak trees on which may be seen a few lengths of ruinous wall. This is the royal castle of Marchmont or Roxburgh which David of Huntingdon founded in the early 12th century as the seat of his principality of Cumbria, which his brother, King Alexander, devolved on him. Later he erected a new burgh of Roxburgh on the haugh land beyond, now used for horse racing. The town grew to be one of the four most important burghs in Scotland in the 13th and 14th centuries and held an annual fair of St James on 25th July.

The castle and burgh arose and fell in response to political requirements. Founded by David I, it was destroyed following the successful siege of 1460 by James II that ejected the English from southern Scotland, killing off both castle and town. The subsequent use of the castle site for a fort in 1547 by protector Somerset was a short-lived episode that came to nothing with the signing of peace in 1550 and the removal of the English garrisons. Subsequently the site has been variously parkland of Friars House belonging to the Earl of Roxburghe in the 17th century and farmland after the move to Floors castle in the 18th century. The last occasion on which the burgh site was ploughed, in the 2nd World War, produced large quantities of medieval pottery. Recent interest has been rekindled by a Time Team excavation in 2003.

The burgh plan has been retrieved from oblivion by a combination of documents, topographical analysis and serendipity. In the hot dry summers of 1984 and 2006 parch marks in the grass revealed the drainage ditches that lined the streets of the burgh, the outlines of buildings with stone footings that fronted on to them and the great ditches that delimited the burgh ramparts. Indeed the streets can in places be observed on the ground where the camber of the roads is still recognisable. This has all been made possible by a 3D survey using high resolution GPS during 2006-9 that covered every part of the burgh and the castle mound. The names of several streets are mentioned in property transactions called charters that survive in monastic records and royal charters. These include King Street, an east-west street, Market Street, a north-south street and Senegede Street. The main east-west street, probably King Street, ran along a natural terrace on the north of the castle mound, and made a junction with a north-south street just south of the Tweed near the east rampart, which may be Market Street. A raid by the earl of Dunbar in 1385 to avenge the killing of one of his men, laid waste the market which lay along it and may be located where the north-south street rides up over a higher terrace. Here in a space where another road meets the north-south street lines of square pits may be seen in the cropmarks which may be the cellars of, or for the rubbish of market booths.

The castle

The castle already housed a strong tower in the reign of King David in which he imprisoned Malcolm, brother of the Mormaer of Moray in 1134 who was foolish enough to raise the standard of revolt against him. No definite trace of the tower now remains although a high point on the castle with a lump of masonry on its north side may be a relic of this strong point. It sits at the south-west corner of a central court of the castle that is shown on a plan of 1547 as ‘The bell mowntte’. This occupies the north-west corner of the artillery fort that was built on the old castle mound during the war of the Rough Wooing to hold down southern Scotland. It was short lived but made a huge impact on the castle. Enough remains to determine that there was one long enclosure that narrows at its north-east end. There is an entrance at each end, ruins of which can still be seen. Of the two that at the south-west is a wide entrance defended by two towers and a yett, wear marks of which can be seen in the left hand tower as you enter. At the north end one side of a barbican with a doorstop still stands above a steep slope down to the great ditch than defends three sides of the castle, leaving the Teviotside open apart from a long wall of masonry at the top of the vertiginous slope of the mound. Two arched gaps in the wall mark the position of postern gates and there is a hole in the wall half way along in which there is a drain for the disposal of liquid rubbish or slops from the castle kitchens. Nearby to the south another gap gives way to a terrace that was once the location of the Captain’s lodging shown on a plan in 1547, but may have been a rebuild of an earlier hall.


Martin, C. and Oram, 2007, R. Medieval Roxburgh: a preliminary assessment of the burgh and its locality. In Proceedings of the Society of Antiquaries of Scotland 137 (2007), pp 357-404.

RCAHMS, (1956), The Royal Commission on the Ancient and Historical Monuments of Scotland. An inventory of the ancient and historical monuments of Roxburghshire: with the fourteenth report of the Commission, 2v. Edinburgh, pp 407-411

Dr Piers Dixon, Deputy Head of Survey and Recording, HES

Sbc Note

Visibility: This is an upstanding earthwork or monument.

Information from Scottish Borders Council


MyCanmore Image Contributions

Contribute an Image

MyCanmore Text Contributions